In order to most effectively interpret variable amplitude fatigue data it is necessary to establish a parameter which characterizes the particular stress history and which permits test results to be correlated with constant amplitude data. In the present research a statistical stress analysis theory is reviewed which permits calculation of an equivalent stress for a realistic (variable amplitude) fatigue loading spectrum, and this has been applied toward interpretation of research data where welded steel specimens were fatigued in flowing natural sea water: Analysis of various alternative equivalent stress parameters has permitted determination of how variable amplitude corrosion fatigue data may best be correlated with constant amplitude results. In addition, the equivalent stress concept has been extended to fracture mechanics analysis in terms of an effective stress intensity; and fatigue data has also been evaluated in terms of this. The resultant correlations indicate good agreement between calculated and experimental results.
Due to the progressive development in recent years of deep water structures, fatigue of welded steel in sea water has become an increasingly important topic (1). At the same time only a limited amount of variable amplitude fatigue testing has been performed under conditions which realistically conform to service loadings at actual welded connections (for example, see reference 2). Such activities have addressed two-primary objectives which are, first, the development of fatigue data for welded steel specimens under conditions of variable amplitude, random-sequence stressing, both with and without cathodic protection, and. second, development of an analytical method for establishing and evaluating such variable amplitude data.
As a component of the above and related project field measurements of wave data and response of offshore structures have been analyzed, and various hypotheses for predicting the fatigue behavior of specimens and welded connections have been proposed and compared. A critical problem in accomplishing this pertains to representation of cycle-to-failure data for variable amplitude (or deflection) fatigue tests in an S-N curve format and comparison of this data with that from constant amplitude experiments. This question becomes particularly critical when one considers that design curves have been based upon constant amplitude or deflection data. The root mean square stress ran has been widely employed for this purpose; and although there is no physical basis for employing this parameter in the application in question, still satisfactory correlations have been realized in fatigue projects pertaining to welded bridge, aircraft and automotive applications.
In the present evaluation the linear cumulative approach for assessing fatigue damage has been adapted. In addition, a statistical stress analysis technique is reviewed for calculation of an equivalent stress. This equivalent stress parameter has been extended to fracture mechanics analysis in terms of an effective stress intensity and to fatigue life estimation.
